Download presentation

Presentation is loading. Please wait.

Published bySavanah Mustin Modified about 1 year ago

1
Online Social Networks and Media Navigation in a small world

2
Small world phenomena Small worlds: networks with short paths Obedience to authority (1963) Small world experiment (1967) Stanley Milgram ( ): “The man who shocked the world”

3
Small world experiment Letters were handed out to people in Nebraska to be sent to a target in Boston People were instructed to pass on the letters to someone they knew on first-name basis The letters that reached the destination followed paths of length around 6 Six degrees of separation: (play of John Guare)

4
Milgram’s experiment revisited What did Milgram’s experiment show? – (a) There are short paths in large networks that connect individuals – (b) People are able to find these short paths using a simple, greedy, decentralized algorithm

5
Small worlds We can construct graphs with short paths – E.g., the Watts-Strogatz model

6
Small worlds Same idea to different graphs

7
Navigation in a small world Kleinberg: Many random graphs contain short paths, but how can we find them in a decentralized way? In Milgram’s experiment every recipient acted without knowledge of the global structure of the social graph, using only – information about geography – their own social connections

8
Kleinberg’s navigation model Assume a graph similar (but not the same!) to that of Watts-Strogatz – There is some underlying “geography”: ring, grid, hierarchy Defines the local contacts of a node Enables to navigate towards a node – There are also shortcuts added between nodes The long-range contacts of a node Similar to WS model – creates short paths

9
Kleinberg’s navigational model Given a source node s, and a navigation target t we want to reach, we assume – No centralized coordination Each node makes decisions on their own – Each node knows the “geography” of the graph They can always move closer to the target node – Nodes make decisions based only on their own contacts (local and long-range) They do not have access to other nodes’ contacts – Greedy (myopic) decisions Always move to the node that is closest to the target.

10
Example

11
Long-range contacts

12
Clustering exponent This exponent is the only one for which we can obtain “short” (polylogarithmic length) paths

13
Theoretical results

14
Proof Intuition The algorithm has the same probability to link to any scale of resolution – logn scales, logn steps in expectation to change scale

15
Proof intuition The algorithm is able to replicate what happens in the Milgram experiment

16
Long range links in the real world

17
Linking by rank

18
Live Journal measurements Replicated for other networks as well (FB) Is there a mechanism that drives this behavior?

19
Other models Lattice captures geographic distance. How do we capture social distance (e.g. occupation)? Hierarchical organization of groups – distance h(i,j) = height of Least Common Ancestor

20
Other models

21
Theorem: For α =1 there is a polylogarithimic search algorithm. For α ≠1 there is no decentralized algorithm with poly-log time – note that α =1 and the exponential dependency results in uniform probability of linking to the subtrees

22
Generalization

23
Doubling dimension

24
Small worlds with nodes of different status

25
Application: P2P search -- Symphony Map the nodes and keys to the ring Link every node with its successor and predecessor Add k random links with probability proportional to 1/(dlogn), where d is the distance on the ring Lookup time O(log 2 n) If k = logn lookup time O(logn) Easy to insert and remove nodes (perform periodical refreshes for the links)

26
Proof of Kleinberg’s theorem

27
Normalization constant

28

29

30

31

Similar presentations

© 2016 SlidePlayer.com Inc.

All rights reserved.

Ads by Google